Abstract: Based on simulations from PIOMAS (Pan-Arctic Ice Ocean Modeling and Assimilation System) model developed by University of Washington and their comparison with observational data, this study analyzes the characteristic dynamic and thermodynamic processes of the decrease in Arctic sea ice. The sea ice volume output flux is used to study dynamic process while thermodynamic process is obtained from dynamics indirectly in the recent 30 years, especially since the 1990s. The correlations between the Arctic sea ice decrease and atmospheric circulation modes like Arctic Oscillation (AO) and Arctic Dipole (AD) are also discussed. Results show that the thickness of perennial sea ice over Fram Strait experienced a significant reduction since 1995. Thermodynamic process played a prominent role in the Arctic sea ice decrease since the 1990s, especially in the 2000s. Furthermore, AO and AD do have influences on the flux of Arctic sea ice volume and their correlations with the flux are all significant at the 99% confidence level. Multiple linear regression using AO and AD can fit the reduction of sea ice volume output flux perfectly.